Lecture 6: Language Production

Question 1

stages of speech production (6)

Answer 1

• 1) Conceptualizing the overall meaning we want to express
• 2) Formulating what words to use, and what order to put them in.
  
  • Lexical access: select words to express these concepts.
  • Organize words syntactically to convey a message.
• 3) Articulation: Implement each word as a series of sounds that comes out of our mouths.
• 4) Monitoring: Repair/stop and restart/utter editing expression (e.g. I’m sorry, I mean).

Question 2

disfluencies in speech planning (5)

Answer 2

• Many types of disfluencies: pauses, breaks, false starts, repairs, repetitions, fillers (e.g. uh, uhm), prolongations, etc.
• Aren’t randomly distributed; occur under heavy cognitive load.
  
  • e.g. 1) Planning a new utterance, 2) describing something difficult, 3) distraction, 4) before low frequency words, 5) before words that have more alternatives, 6) before objects that don’t have a conventional name, etc.
• Listeners are sensitive to speakers’ difficulty.
  
  • e.g. Participants shown an ice cream cone and a weird shape, then hear Click on the uh… → look more to the weird shape that’s harder to describe.

Question 3

What do speech errors tell us about production? (2)

Answer 3

• What the planning units are.
• When different parts are being planned.

Question 4

word-level speech errors (6)

Answer 4

• substitution; e.g. I put the mushrooms in the water. (I put the flowers in the water.)
• exchange; e.g. write a mother to my letter
• anticipation; e.g. sky is in the sky
• Errors almost always involve two words of the same syntactic category.
  
  • e.g. Once I stop, I can’t start. You would insert a noun here.
• They also tend to involve two words that are semantically connected.

Question 5

morpheme-level speech errors (3)

Answer 5

• When we insert morphemes in the wrong place.
• addition; e.g. wanted to strained it.
• shift; e.g. I want to readed (I wanted to read)

Question 6

phoneme-level speech errors (2)

Answer 6

• perseveration; beef needle (beef noodle)
• Happens most with long vowels.

Question 7

lexical bias (1)

Answer 7

• The statistical tendency for sound-based speech errors to result in actual words rather than nonwords.

Question 8

Baars, Motley, & McKay (1975) (4)

Answer 8

• Had subjects read word pairs.
  
  • darn boore → barn door
  • dart board → bart board
• Were more likely to produce the first error.
• Demonstrated the lexical bias.

Question 9

types of word-level errors (4)

Answer 9

• phonological errors; e.g. cat → mat
• semantic errors; e.g. cat → dog
• ‘mixed’ errors; e.g. cat → rat
  
  • More likely than cat → mat and cat → dog together.

Question 10

How is the lexical entry accessed? (3)

Answer 10

• Option 1: The whole lexical entry is accessed; bag of marbles metaphor: you just pick everything out of a bag of marbles.
• Option 2: We access meaning first, sound second.
  
  • What we see evidence for.

Question 11

two-step model (1)

Answer 11

• Message –Lexicon→ Lemma Level (Grammatical Encoding) → Lexeme Level (Phonological Encoding) –Lexicon→ Articulation

Question 12

picture-word interference task (7)

Answer 12

• Provides evidence for the two-step model.
• Participants see a picture (the target; e.g. a dog) with a word on it (the prime).
  
  • Primes can be: unrelated (e.g. dog-bed), semantically related (e.g.dog-cat), or phonologically related (e.g. dog-doll).
• Dependent variable: latency to name picture.
  
  • Semantic relation → slowdown; interference.
  • Phonological relation → facilitation.
• Provides indirect evidence that meaning comes before sound.

Question 13

Meyer (1996) (6)

Answer 13

• Study conducted with Dutch participants; had to say sentences with the structure The X is next to Y.
  
  • e.g. The church is next to the arrow.
• Experiment 1: Before hearing target sentence, hear a distractor word that’s semantically related/unrelated to either X or Y.
  
  • Result: Semantic interference; reaction times to both words (X and Y) were slower than when the distractor word was unrelated.
• Experiment 2: Before hearing the target sentence, participants hear a distractor word that’s phonologically related/unrelated to either X or Y.
  
  • Result: Phonological facilitation effects for only the first word; even if the distractor word rhymed with the second word, there was no effect.
• Conclusion: Meaning comes before phonology; retrieval of meaning “lemma” happens while the utterance is being planned, but the planning of sound comes later—while uttering the sentence.

Question 14

What affects the order/structure of the sentence? (5)

Answer 14

• The message to be conveyed.
• The grammar of one’s language.
• Subtle shadings of meaning or emphasis.
  
  • e.g. The garden swarmed with bees vs. Bees swarmed in the garden.
  • Both of these communicate the same relations, but the first sentence seems to suggest (more than the second) that the garden is mostly filled with bees.
• Lexical selection; heavy-NP shift.

Question 15

heavy-NP shift (1)

Answer 15

• A syntactic structure in which a long NP, usually a direct object, is moved toward the end of the sentence instead of in its normal spot adjacent to the verb.

Question 16

Ferreira (1996) (6)

Answer 16

• Participants created sentences out of collections of words that were presented on a screen.
• These verbs allowed participants to either create one or two versions of a sentence.
  
  • e.g. For the verb gave: I gave the children the toys or I gave the toys to the children.
  • e.g. For the verb donated: I donated the toys to the children. It sounds strange to say I donated the children the toys.
• Participants made fewer errors and spoke faster when they had to create a sentence using the more flexible verb gave.
• Reveals that speakers use syntax as a way of easing the burden of speaking.

Question 17

Brock (1986a) (5)

Answer 17

• Participants presented with a picture of a church being struck by lightning and asked to describe the picture.
• Primed with thunder (semantically related to lightning) or worship (semantically related to church).
• If they heard thunder, they were more likely to use the active structure (73% vs. 12%), but if they were primed with worship they were more likely to use the passive structure (20% vs. 65%).
  
  • Lightning is striking the church (active) vs. The church is being struck by lightning (passive).
• Note that there wasn’t a complete reversal; participants were still used the active structure a majority of the time, but the result is still reliable.

Question 18

Why are active sentences more commonly produced? (4)

Answer 18

• Frequency? Syntactic complexity? Animacy? Likely the latter.
• Preference for an animate ‘agent’ subject, and an inanimate ‘patient’ object.
  
  • In a passive, the subject is the patient, and often, inanimate.
  • We try to avoid this mismatch between syntactic role and semantic role (i.e. we want the subject to be animate and the object to be inanimate).

Question 19

Brock (1986b) (6)

Answer 19

• Participants saw a picture and told to describe it. Before that, they heard a sentence and had to repeat it (prime).
  
  • e.g. Prepositional sentence: A rock star sold some cocaine to an agent.
  • e.g. Double object sentence: A rock star sold an agent some cocaine.
• Primed with prepositional structure → 48% prepositional, 31% double object.
• Primed with double object structure → 25% prepositional, 53% double object.
• Showed that the double object structure is preferred overall, but there is still an effect of priming.